// Remove outlines that are a tiny fraction in either width or height
// of the word height.
void Textord::clean_small_noise_from_words(ROW *row) {
WERD_IT word_it(row->word_list());
for (word_it.mark_cycle_pt(); !word_it.cycled_list(); word_it.forward()) {
WERD* word = word_it.data();
int min_size = static_cast<int>(
textord_noise_hfract * word->bounding_box().height() + 0.5);
C_BLOB_IT blob_it(word->cblob_list());
for (blob_it.mark_cycle_pt(); !blob_it.cycled_list(); blob_it.forward()) {
C_BLOB* blob = blob_it.data();
C_OUTLINE_IT out_it(blob->out_list());
for (out_it.mark_cycle_pt(); !out_it.cycled_list(); out_it.forward()) {
C_OUTLINE* outline = out_it.data();
outline->RemoveSmallRecursive(min_size, &out_it);
}
if (blob->out_list()->empty()) {
delete blob_it.extract();
}
}
if (word->cblob_list()->empty()) {
if (!word_it.at_last()) {
// The next word is no longer a fuzzy non space if it was before,
// since the word before is about to be deleted.
WERD* next_word = word_it.data_relative(1);
if (next_word->flag(W_FUZZY_NON)) {
next_word->set_flag(W_FUZZY_NON, false);
}
}
delete word_it.extract();
}
}
}
// If this outline is smaller than the given min_size, delete this and
// remove from its list, via *it, after checking that *it points to this.
// Otherwise, if any children of this are too small, delete them.
// On entry, *it must be an iterator pointing to this. If this gets deleted
// then this is extracted from *it, so an iteration can continue.
void C_OUTLINE::RemoveSmallRecursive(int min_size, C_OUTLINE_IT* it) {
if (box.width() < min_size || box.height() < min_size) {
ASSERT_HOST(this == it->data());
delete it->extract(); // Too small so get rid of it and any children.
} else if (!children.empty()) {
// Search the children of this, deleting any that are too small.
C_OUTLINE_IT child_it(&children);
for (child_it.mark_cycle_pt(); !child_it.cycled_list();
child_it.forward()) {
C_OUTLINE* child = child_it.data();
child->RemoveSmallRecursive(min_size, &child_it);
}
}
}